Of the 120 million patients seen annually in US Emergency Departments, 85% are discharged home, as are the vast majority of the one billion patients seen in ambulatory clinics. While rates of in-hospital deaths are routinely measured and reported, deaths outside of health facilities are far more challenging to capture because of fragmented health information systems both within and among health care organizations. As a result, providers have no systematic way of knowing when a patient dies unexpectedly after being sent home, and policy makers cannot readily measure how frequently these catastrophic events occur. I propose a series of linked research projects to track and analyze unexpected deaths after medical encounters, using data from a large academic hospital system as well as Medicare claims, in order to identify factors related to individual patients, providers, and system that increase the likelihood of such events. I will subsequently adapt the methods to an international setting, where clinical and administrative data are less complete than in the US. This work would have important implications for improving both clinical care and health services design. In addition, a systematic method for tracking and reporting rates of unexpected early deaths would be a useful way to measure health services performance, and assess trade-offs between quality and cost.